Project Summary: In the previous cycle of this grant, we characterized hemisphere-specific motor control deficits in the non- paretic arm of unilaterally lesioned stroke survivors. Our preliminary data indicate these deficits are substantial and functionally limiting in patients with severe paresis. We have specifically designed an intervention to remediate the hemisphere-specific deficits in the non-paretic arm, using a virtual-reality platform, and then follow this training with manipulation training of a variety of real objects, designed to facilitate generalization and transfer to functional behaviors encountered in the natural environment. We propose a 2-site, two-group randomized intervention with a treatment group, which will receive unilateral training of the non-paretic arm, through our Virtual Reality and Manipulation Training (VRMT) protocol. This intervention protocol is grounded in the premise that targeted remediation of fundamental control deficits exhibited by the non-paretic arm will generalize and transfer beyond practiced tasks to performance of activities of daily living (ADL). This approach contrasts with the more pragmatic approach of task-specific training of essential ADL?s, which is limited in scope, more cumbersome, and ignores known fundamental motor control deficits. Our control group will receive conventional intervention, guided by recently released practice guidelines for upper limb intervention in adult stroke. The impact of the proposed research is that we address persistent functional performance deficits in chronic stroke patients with severe paresis, who?s non- paretic arm impairments are generally ignored in most current rehabilitation protocols. Our first aim addresses the overall effectiveness of this intervention, relative to our control group: To determine whether non-paretic arm VRMT in chronic stroke survivors with severe paresis will produce durable improvements in non-paretic arm motor performance that will generalize to improve functional activities and functional independence to a greater extent than conventional therapy focused on the paretic arm. Our second aim focuses on the mechanistic basis of potential training-related improvements in motor performance: To determine whether intervention-induced improvements in non-paretic arm performance are associated with improvements in hemisphere-specific reaching kinematics. Finally, our third aim monitors for potential negative effects of our experimental intervention on paretic arm impairment. We have already integrated the PI?s laboratories (Sainburg-PSU, Winstein-USC) for our pilot research project that has provided excellent support for aims 1 and 3. !